Leveraging the capabilities provided by their deformable bodies, continuum soft robots promise to have a disruptive impact on several fields of science, industry, and society. In the past decade, substantial progress has been made in developing soft robotic platforms, but significantly less attention has been devoted to the control challenge. Today, soft robotic systems still need a low-level artificial brain that can make them execute precise motions and eventually exploit the intelligence embedded in their complex mechanical structures. In this talk, I will briefly introduce this grand challenge within the soft robotic field and then introduce the model-based view of its solution. I will show how simplified models can be combined with nonlinear control theory, leading to precise and dynamic task execution. I will then discuss how these techniques can be combined with data and machine learning.